Modeling of traffic queue discharge dynamics

Active: 
yes

The goal of the project is to develop a mathematical model (or models) of dynamics of discharge of a queue of road vehicles at intersections controlled by traffic lights. The model(s) will be used in later projects for design of feedback control algorithms for traffic lights, in particular the preemption algorithms based on communication between emergency vehicles (ambulance, police, fire trucks) and the traffic lights controller (see the diploma thesis by Vít Obrusník and his video from real traffic).

The existing infrastructure, in particular the induction loop detectors and road-side cameras are not accessible to us in this project, therefore our own instrumentation will have to be designed and implemented as the first step in the project. Most probably it will be based purely on digital cameras and subsequent image recognition. An expexted outcome is a set of recorded speed profilles/trajectories of all the vehicles in the queue after the traffic lights turn green. One or perhaps several digital cameras could be installed temporarily (for just a few minutes) at some chosen intersection to acquire the data/images/videos, from which the motion of individual vehicles, hence their speeds, will be extracted.

The recorded speed profiles will be analyzed and several basic vehicle-following models will be used to tried to "explain" the data. The work has a potential to result in a research publication because the currently known car-following models (such as Intelligent Driver Model) are mainly used in highways and it is not obvious is the dynamics at intersections differs only in the parameters or the whole structure.

The topic is suitable for a team project: finding a suitable spot (possibly a window on some higher floor of a building next to an intersection), selecting and setting up the camera(s), extracting the "features" from the recorded videos, car-following modeling.

The project will be co-supervised by prof. Karel Zimmerman from Center for Machine Perception (CMP).

Contact person: 
Zdeněk Hurák
Image: Traffic lights
Image: Traffic lights at an intersection